Fouling in plate-and-frame heat exchangers (PHEs) may be defined as the deposition of unwanted material on the heat transfer surface that reduces heat-transfer and increases the resistance to fluid flow. Once the thermal-hydraulic performance decreases to a minimum acceptable level, cleaning of the equipment has to be done to restore the performance. The decision regarding periodic off-line maintenance of the heat exchangers is generally based on a thermal-economic performance of the process. In this paper, we discuss a probabilistic maintenance model for PHE by incorporating the risk level and scatter parameter of the four random fouling growth models, namely linear, power law, falling rate and asymptotic models, which are integrated in the dimensionless cost model for a heat exchanger used in a steel plant. All the results are presented in terms of non-dimensional plots. The results show that there is a strong relationship between t(down) and the uptime, particularly in the region where the costs of operation and maintenance are minimum. Copyright (c) 2005 John Wiley & Sons, Ltd.